Your 23andMe raw data file, what it actually contains and what to do with it

If you took a 23andMe test, you paid for two things. The first is the curated reports they show you in your account, ancestry breakdown, traits, a small handful of health markers. The second is the raw genetic data file, downloadable as a single text file from your account settings.

Most users never download it. The ones who do often have no idea what's in it or what to do with it. That's a shame, because the raw file is by far the more valuable artifact. The reports are a curated slice. The raw file is the data.

This post walks through exactly what's in that file, what it actually means, and what your honest options are now that you have it. Think of it as a primer on how to read a DNA raw data file, written for people who never plan to become geneticists.

What 23andMe (and AncestryDNA, MyHeritage) actually measure

These tests do not sequence your whole genome. They use a technique called SNP genotyping, they pick about 600,000 to 700,000 specific positions in your genome where humans are known to vary, and they read which letter you carry at each one.

For context: your whole genome has about 3 billion letters. A consumer test reads roughly 0.02% of them, focused on the positions where the difference between people is most informative.

The raw file you can download is a list of those positions, with three pieces of information per row:

• The SNP identifier (rs429358, rs1801133, etc.), the same rs IDs cataloged in NCBI's dbSNP database
• The chromosome and position
• Your genotype, the two letters you carry at that position (one from each parent)

A typical row looks like:

`

rs9939609 16 53820527 AT

`

That's the SNP rs9939609, on chromosome 16, position 53820527, where you carry one A allele and one T allele.

Multiply that by 600,000+ rows and you have your file. That is the 23andMe txt file meaning in one sentence: a long list of (SNP, chromosome, position, genotype) tuples.

What you can do with this file

This is where the consumer products often miss. Once you have the raw file, you have everything you need to look up what published research says about any of those 600,000+ positions. The bottleneck is interpretation, not data.

There are essentially four things you can do.

1. Upload it to another interpretation service. Several products exist that interpret 23andMe raw data and surface what the research says about each variant. Privacy posture varies a lot. So does evidence quality. Promethease, the longtime favorite, wound down operations. Genetic Lifehacks, SelfDecode, Genomelink are alternatives at different price points and editorial styles. Expressive (us) is the privacy-first DNA analysis option, with evidence quality always visible on the page.

2. Look up specific variants yourself. If you only care about a handful of well-studied SNPs (APOE for Alzheimer's risk, MTHFR for folate metabolism, HLA-B*57:01 for one specific drug sensitivity), you can analyze 23andMe raw data by hand: look up your genotype against the published research at PubMed, ClinVar, the NHGRI-EBI GWAS Catalog, or PharmGKB directly. This is the most thorough option but requires the most work per variant.

3. Use it for clinical context. If you ever get prescribed a medication that has a pharmacogenomic flag, warfarin, clopidogrel, certain SSRIs, abacavir, your raw file probably contains the variants that determine how well the drug will work for you. Bring it to the prescribing physician.

4. Leave it on your computer. Many people just want to know they have it, in case future interpretations get better. That's a legitimate choice. The file is small (about 25 MB compressed), portable, and doesn't go stale, the underlying biology is what it is.

Why the raw file matters more than the curated reports

The curated reports, the ones 23andMe builds for you in their app, are limited by three constraints:

• FDA regulatory caution. 23andMe can only report on health markers the FDA has explicitly authorized. The list is short, and it's heavily biased toward variants that are unambiguously interpretable. Anything ambiguous or "interesting but not clinically actionable" gets left out.
• Editorial caution. Once a variant is reported, 23andMe takes liability for the interpretation. The result is wording so carefully hedged that the actual takeaway often gets lost.
• Sparsity by design. The consumer product can only show you the variants 23andMe chose to highlight. Your raw file contains hundreds of thousands more, including pharmacogenomic markers, eQTLs, and trait associations that 23andMe doesn't surface.

The raw file is the part you actually paid for. The reports are the marketing surface.

How to download it

This changes occasionally as 23andMe redesigns their site, but as of mid-2026:

1. Log in to 23andMe.
2. Go to your name then Settings.
3. Scroll to "Raw Data" or "Browse Raw Data."
4. Click the download button. They'll send you a verification email; click the link to confirm.
5. You'll get a .zip file. Inside is a plain-text file (or .txt) that's your raw data.

23andMe's own customer care article on raw data download covers the current click path if their UI has moved since this post went up.

If you're an AncestryDNA customer, the process is broadly similar: Account, then DNA, then Download Raw DNA Data. MyHeritage has it under DNA, then Manage DNA Kits.

Save it somewhere you control. Keep a backup. Once you've downloaded it, the file is yours, independent of whether you keep your 23andMe account. That is the whole point of data sovereignty in this context: your genome stays yours, on your machine, on your terms.

What about privacy?

The raw file you download is local. Nothing on your computer phones home to 23andMe. So long as the file stays on your machine, your data is exactly as private as your computer.

Where things change is when you upload the file to another service to interpret it. Every service has a different privacy posture. Some store the file on their servers, indefinitely, sometimes with the right to use it for research. Some process it client-side in your browser and discard the raw bytes immediately. Most are in between.

We built Expressive specifically to make the privacy story explicit: AES-256-GCM encryption with keys derived from your wallet signature, HMAC-only lookups so the system can find your records without ever decrypting them server-side, no marketing pixel firing on your genome. We talk about our technical handling here.

But you don't have to pick us. The general principle is: read the privacy policy of whichever service you upload to, ignore the marketing copy, look at the technical specifics. If the company can describe their encryption model in one sentence and the sentence makes sense, that's a good sign. If they describe it in vague generalities, it usually means there isn't one.

What to do next

Download the file. Even if you don't do anything with it today, you'll have it forever. The next interpretation service you hear about will be a one-click upload away.

If you're ready to interpret it now, Expressive accepts raw genetic files for 23andMe, AncestryDNA, MyHeritage, Sequencing.com, and standard VCF format. We anchor every claim to a peer-reviewed study, we tell you when the evidence is thin, and we never sell your data, because we built it so we technically can't.

Want updates when we ship new variant pages or a research deep-dive? Read the latest issue or get notified about early access.